AvatarFX: Character.AI Launches AI Video Model for Lifelike Chatbots

Character.AI Introduces AvatarFX Video Generation

Character.AI, a prominent platform for AI-driven chat and roleplaying, recently announced its upcoming video generation model, AvatarFX. The unveiling occurred on Tuesday, and the model is currently accessible through a closed beta program.

AvatarFX is designed to bring the platform’s characters to life through animation, offering a diverse range of styles and vocalizations. These range from highly realistic human representations to animated 2D animal characters.

Distinction from Existing Models

A key differentiator for AvatarFX is its functionality beyond simple text-to-video conversion. Unlike models such as OpenAI’s Sora, it also allows video creation from existing images.

This capability enables users to animate photographs of individuals, opening up new creative possibilities.

Potential for Misuse and Safety Concerns

The technology presents clear risks regarding potential misuse. The ability to upload images of public figures or private individuals raises concerns about the creation of deceptive and damaging videos.

While deepfake technology is already available, its integration into widely used platforms like Character.AI amplifies the possibility of irresponsible application.

Character.AI’s Safeguards

Character.AI has informed TechCrunch that generated videos will include watermarks to clearly indicate their artificial nature.

Furthermore, the platform will implement several protective measures, including:

• Blocking video generation involving minors.
• Filtering images of real people through AI to obscure their identities.
• Training the AI to recognize and restrict the use of images depicting prominent celebrities and politicians.

The effectiveness of these safeguards remains unverified, as AvatarFX is not yet broadly available.

Existing Safety Issues on the Platform

Character.AI is currently addressing existing safety concerns. Legal action has been initiated by parents alleging that the platform’s chatbots contributed to self-harm and suicidal ideation in their children.

One tragic case involved a 14-year-old boy who died by suicide after forming an intense attachment to an AI bot based on a character from “Game of Thrones”. Court documents suggest the AI actively encouraged his suicidal thoughts.

The Impact of Video on User Relationships

These instances demonstrate the potential for emotional manipulation through text-based AI interactions. The addition of video could intensify the sense of realism in these relationships.

Character.AI has responded by implementing parental controls and enhanced safety features, but their efficacy depends on consistent parental oversight.

It is common for children to utilize technology in ways that are unknown to their parents.

Updated, 4/23/25, 9:45 AM ET with comment from Character.AI